论文检索

General Machine Learning · Probabilistic Methods

Puhua Niu, Shili Wu, Xiaoning Qian

Generative Flow Networks (GFlowNets) were developed to learn policies for efficiently sampling combinatorial candidates by interpreting their generative processes as trajectories in directed acyclic graphs. In the value-based training workflow, the objective is to enforce the balance over partial episodes between the flows of the learned policy and the estimated flows of the desired policy, implicitly encouraging policy divergence minimization. The policy-based strategy alternates between estimating the policy divergence and updating the policy, but reliable estimation of the divergence under directed acyclic graphs remains a major challenge. This work bridges the two perspectives by showing that flow balance also yields a principled policy evaluator that measures the divergence, and an evaluation balance objective over partial episodes is proposed for learning the evaluator. As demonstrated on both synthetic and real-world tasks, evaluation balance not only strengthens the reliability of policy-based training but also broadens its flexibility by seamlessly supporting parameterized backward policies and enabling the integration of offline data-collection techniques. Our code is available at [github.com/niupuhua1234/Sub-EB](https://github.com/niupuhua1234/Sub-EB).

General Machine Learning · Representation Learning

András Balogh, Márk Jelasity

Generative sequence models are typically trained on sample sequences from natural or formal languages. It is a crucial question whether—or to what extent—sample-based training is able to capture the true structure of these languages, often referred to as the "world model". Theoretical results indicate that we can hope for soundness at best, that is, generating valid sequences, but not necessarily all of them. However, it is still important to have practical tools that are able to verify whether a given sequence model is sound. In this study, we focus on chess, as it is a domain that provides enough complexity while having a simple rule-based world model. We propose adversarial sequence generation for verifying the soundness of the sequence model. Our adversaries generate valid sequences so as to force the sequence model to generate an invalid next move prediction. Apart from the falsification of soundness, this method is also suitable for a more fine-grained analysis of the failure modes and the effects of different choices during training. To demonstrate this, we propose a number of methods for adversarial sequence generation and evaluate the approach on a large set of chess models. We train models on random as well as high-quality chess games, using several training recipes. We find that none of the models are sound, but some training techniques and dataset choices are able to improve soundness remarkably. We also investigate the potential application of board state probes in both our training and attack methods. Our findings indicate that the extracted board states have no causal role in next token prediction in most of the models.

General Machine Learning · Probabilistic Methods

Yang Yang, Severi Rissanen, Paul Chang, Nasrulloh Satrio, Daolang Huang, Arno Solin, Markus Heinonen, Luigi Acerbi

Amortized simulator-based inference offers a powerful framework for tackling Bayesian inference in computational fields such as engineering or neuroscience, increasingly leveraging modern generative methods like diffusion models to map observed data to model parameters or future predictions. These approaches yield posterior or posterior-predictive samples for new datasets without requiring further simulator calls after training on simulated parameter-data pairs. However, their applicability is often limited by the prior distribution(s) used to generate model parameters during this training phase. To overcome this constraint, we introduce *PriorGuide*, a technique specifically designed for diffusion-based amortized inference methods. PriorGuide leverages a novel guidance approximation that enables flexible adaptation of the trained diffusion model to new priors at test time, crucially without costly retraining. This allows users to readily incorporate updated information or expert knowledge post-training, enhancing the versatility of pre-trained inference models.

General Machine Learning · Representation Learning

Paul Youssef, Zhixue Zhao, Christin Seifert, Jörg Schlötterer

Knowledge editing methods (KEs) are a cost-effective way to update the factual content of large language models (LLMs), but they pose a dual-use risk. While KEs are beneficial for updating outdated or incorrect information, they can be exploited maliciously to implant misinformation or bias. In order to defend against these types of malicious manipulation, we need robust techniques that can reliably detect, interpret, and mitigate malicious edits. To that end, we introduce the tasks of tracing and reversing edits. We propose a novel method to infer the edited object entity, solely based on the modified weights, without access to the editing prompt or any other semantically similar prompts, with up to 99\% accuracy. Further, we propose an effective and training-free method for reversing edits. Our method reverses up to 94\% of the edits, and helps regain the original model's output distribution without access to any information about the edit. This method can further be repurposed to distinguish between edited and unedited weights. Our findings highlight the feasibility of tracing and reversing edits based on the edited weights, opening a new research direction for safeguarding LLMs against adversarial manipulations.

General Machine Learning · Transfer, Multitask and Meta-learning

Siqi Wang, Aoming Liu, Bryan Plummer

Methods addressing Learning with Noisy Labels (LNL) and multi-source Domain Generalization (DG) use training techniques to improve downstream task performance in the presence of label noise or domain shifts, respectively. Prior work often explores these tasks in isolation, and the limited work that does investigate their intersection, which we refer to as Noise-Aware Generalization (NAG), only benchmarks existing methods without also proposing an approach to reduce its effect. We find that this is likely due, in part, to the new challenges that arise when exploring NAG, which does not appear in LNL or DG alone. For example, we show that the effectiveness of DG methods is compromised in the presence of label noise, making them largely ineffective. Similarly, LNL methods often overfit to easy-to-learn domains as they confuse domain shifts for label noise. Instead, we propose Domain Labels for Noise Detection (DL4ND), the first direct method developed for NAG which uses our observation that noisy samples that may appear indistinguishable within a single domain often show greater variation when compared across domains. We find DL4ND outperforms DG and LNL methods, including their combinations, even when simplifying the NAG challenge by using domain labels to isolate domain shifts from noise. Performance gains up to 12.5% over seven diverse datasets with three noise types demonstrates DL4ND’s ability to generalize to a wide variety of settings.

General Machine Learning · Probabilistic Methods

Kaiyu Li, Yiming Yang, Xiaoyuan Cheng, Yi He, Zhuo Sun

Control variates are variance reduction techniques for Monte Carlo estimators. They play a critical role in improving Monte Carlo estimators in scientific and machine learning applications that involve computationally expensive integrals. We introduce \emph{multilevel control functionals} (MLCFs), a novel and widely applicable extension of control variates that combines non-parametric Stein-based control variates with multi-fidelity methods. We show that when the integrand and the density are smooth, and when the dimensionality is not very high, MLCFs enjoy a faster convergence rate. We provide both theoretical analysis and empirical assessments on differential equation examples, including Bayesian inference for ecological models, to demonstrate the effectiveness of our proposed approach. Furthermore, we extend MLCFs for variational inference, and demonstrate improved performance empirically through Bayesian neural network examples.

Social Aspects · Trustworthy Machine Learning

Naufal Suryanto, Muzammal Naseer, Pengfei Li, Syed Talal Wasim, Jinhui Yi, Juergen Gall, Paolo Ceravolo, Ernesto Damiani

Cybersecurity operations demand assistant LLMs that support diverse workflows without exposing sensitive data. Existing solutions either rely on proprietary APIs with privacy risks or on open models lacking domain adaptation. To bridge this gap, we curate 11.8B tokens of cybersecurity-focused continual pretraining data via large-scale web filtering and manual collection of high-quality resources, spanning 28.6K documents across frameworks, offensive techniques, and security tools. Building on this, we design an agentic augmentation pipeline that simulates expert workflows to generate 266K multi-turn cybersecurity samples for supervised fine-tuning. Combined with general open-source LLM data, these resources enable the training of RedSage, an open-source, locally deployable cybersecurity assistant with domain-aware pretraining and post-training. To rigorously evaluate the models, we introduce RedSage-Bench, a benchmark with 30K multiple-choice and 240 open-ended Q\&A items covering cybersecurity knowledge, skills, and tool expertise. RedSage is further evaluated on established cybersecurity benchmarks (e.g., CTI-Bench, CyberMetric, SECURE) and general LLM benchmarks to assess broader generalization. At the 8B scale, RedSage achieves consistently better results, surpassing the baseline models by up to +5.59 points on cybersecurity benchmarks and +5.05 points on Open LLM Leaderboard tasks. These findings demonstrate that domain-aware agentic augmentation and pre/post-training can not only enhance cybersecurity-specific expertise but also help to improve general reasoning and instruction-following. Project page: https://risys-lab.github.io/RedSage/

General Machine Learning · Transfer, Multitask and Meta-learning

Gregory De Ruyter, Mathias Verbeke, Hans Hallez

Neural Architecture Search (NAS), particularly gradient-based techniques, has proven highly effective in automating the design of neural networks. Recent work has extended NAS to hardware-aware settings, aiming to discover architectures that are both accurate and computationally efficient. Many existing methods integrate hardware metrics into the optimization objective as regularization terms, which introduces differentiability requirements and hyperparameter tuning challenges. This can either result in overly penalizing resource-intensive architectures or architectures failing to meet the hardware constraints of the target device. To address these challenges, we propose ConNAS, a novel gradient-based NAS framework that enforces hardware constraints directly through gradient modification. This approach eliminates the need for differentiable hardware metrics and regularization weights. The novelty in ConNAS lies in modifying gradients with respect to architectural choices, steering the search away from infeasible architectures while ensuring constraint satisfaction. Evaluations on the NATS-Bench benchmark demonstrate that ConNAS consistently discovers architectures that meet the imposed hardware constraints while achieving performance within just 0.14% of the optimal feasible architecture. Additionally, in a practical deployment scenario, ConNAS outperforms handcrafted architectures by up to 1.55% in accuracy under tight hardware budgets. Our code is publicly available at https://gitlab.kuleuven.be/m-group-campus-brugge/distrinet_public/connas.

Social Aspects · Trustworthy Machine Learning

Thibaud Gloaguen, Mark Vero, Robin Staab, Martin Vechev

Finetuning open-weight Large Language Models (LLMs) is standard practice for achieving task-specific performance improvements. Until now, finetuning has been regarded as a controlled and secure process in which training on benign datasets leads to predictable behaviors. In this paper, we demonstrate, for the first time, that an adversary can create compromised LLMs that are performant and benign, yet exhibit adversarial behaviors once finetuned by downstream users. To this end, we propose an attack, FAB (Finetuning-activated Adversarial Behaviors), which compromises an LLM via meta-learning techniques that simulate downstream finetuning, explicitly optimizing for the emergence of adversarial behaviors in the finetuned models. At the same time, the compromised LLM is regularized to retain general capabilities and to exhibit no adversarial behaviors prior to finetuning. As a result, when users finetune (e.g., instruction-tuning, distillation, DPO) the seemingly benign model on their own datasets, they unknowingly trigger its dormant adversarial behavior. We experimentally demonstrate the effectiveness of FAB across multiple LLMs and three commonly considered target behaviors: unsolicited advertising, jailbreakability, and over-refusal. We show that FAB-triggers are robust to various finetuning choices made by the user (e.g., dataset, number of steps, scheduler, post-training algorithm). Our findings challenge prevailing assumptions on the security of finetuning, revealing a critical attack vector.

General Machine Learning · Transfer, Multitask and Meta-learning

Yiyun Zhou, Jingwei Shi, Mingjing Xu, Zhonghua Jiang, Jingyuan Chen

Knowledge Distillation (KD) has emerged as a powerful technique for model compression, enabling lightweight student networks to benefit from the performance of redundant teacher networks. However, the inherent capacity gap often limits the performance of student networks. Inspired by the expressiveness of pretrained teacher networks, a compelling research question arises: is there a type of network that can not only inherit the teacher’s structure but also maximize the inheritance of its knowledge? Furthermore, how does the performance of such an inheriting network compare to that of student networks, all benefiting from the same teacher network? To further explore this question, we propose InherNet, a neural network inheritance method that performs asymmetric low-rank decomposition on the teacher’s weights and reconstructs a lightweight yet expressive network without significant architectural disruption. By leveraging Singular Value Decomposition (SVD) for initialization to ensure the inheritance of principal knowledge, InherNet effectively balances depth, width, and compression efficiency. Experimental results across unimodal and multimodal tasks demonstrate that InherNet achieves higher performance compared to student networks of similar parameter sizes. Our findings reveal a promising direction for future research in efficient model compression beyond traditional distillation.

Theory · Optimization

Travis Gibson

Gradient based optimization algorithms deployed in Machine Learning (ML) applications are often analyzed and compared by their convergence rates or regret bounds. While these rates and bounds convey valuable information they don’t always directly translate to stability guarantees. Stability and similar concepts, like robustness, will become ever more important as we move towards deploying models in real-time and safety critical systems. In this work we build upon the results in Gaudio et al. 2021 and Moreu & Annaswamy 2022 for gradient descent with second order dynamics when applied to explicitly time varying cost functions and provide more general stability guarantees. These more general results can aid in the design and certification of these optimization schemes so as to help ensure safe and reliable deployment for real-time learning applications. We also hope that the techniques provided here will stimulate and cross-fertilize the analysis that occurs on the same algorithms from the online learning and stochastic optimization communities.

General Machine Learning · Causality

Shuai Yang, Qi Yang, Luoxi Tang, Yuqiao Meng, Nancy Guo, Jeremy Blackburn, Zhaohan Xi

Counterfactual reasoning has emerged as a crucial technique for generalizing the reasoning capabilities of large language models (LLMs). By generating and analyzing counterfactual scenarios, researchers can assess the adaptability and reliability of model decision-making. Although prior work has shown that LLMs often struggle with counterfactual reasoning, it remains unclear which factors most significantly impede their performance across different tasks and modalities. In this paper, we propose a decompositional strategy that breaks down the counterfactual generation from causality construction to the reasoning over counterfactual interventions. To support decompositional analysis, we investigate 11 datasets spanning diverse tasks, including natural language understanding, mathematics, programming, and vision-language tasks. Through extensive evaluations, we characterize LLM behavior across each decompositional stage and identify how modality type and intermediate reasoning influence performance. By establishing a structured framework for analyzing counterfactual reasoning, this work contributes to the development of more reliable LLM-based reasoning systems and informs future elicitation strategies.

Zehong Wang, Chuxu Zhang, Jundong Li, Nitesh V. Chawla, Yanfang Ye 0001

Foundation models have revolutionized machine learning by enabling general-purpose reasoning across diverse tasks and domains. These models, pretrained on large-scale data, demonstrate strong adaptability with minimal task-specific supervision, leading to breakthroughs in natural language processing and computer vision. Inspired by this paradigm, Graph Foundation Models (GFMs) have emerged to extend the benefits of foundation models to graph-structured data, which is pretrained on massive graphs and can be fast adapted to different downstream tasks. In this paper, we provide a comprehensive survey of the state-of-the-art techniques of graph foundation models. In particular, we (1) formally categorize the challenges in designing graph foundation models; (2) comprehensively review the existing and recent advances of graph foundation models; (3) extend the graph foundation models in real-world problems; and (4) elucidate open questions and future research directions. Our systematic review summarizes representative models, highlights key design principles, and provides comparative analyses. This survey introduces major topics within foundation models and offers a guide to a new frontier of graph learning. Our extended survey is available at https://arxiv.org/abs/2505.15116.

John Paparrizos, Paul Boniol, Qinghua Liu, Themis Palpanas

Recent advances in data collection technology, accompanied by the ever-rising volume and velocity of streaming data, underscore the vital need for time series analytics. In this regard, time-series anomaly detection has been an important activity, entailing various applications in fields such as cyber security, financial markets, law enforcement, and health care. While traditional literature on anomaly detection is centered on statistical measures, the increasing number of machine learning algorithms in recent years call for a structured, general characterization of the research methods for time-series anomaly detection. In this paper, we present a process-centric taxonomy for time-series anomaly detection methods, systematically categorizing traditional statistical approaches and contemporary machine learning techniques. Beyond this taxonomy, we conduct a meta-analysis of the existing literature to identify broad research trends. Given the absence of a one-size-fits-all anomaly detector, we also introduce emerging trends for time-series anomaly detection. Furthermore, we review commonly used evaluation measures and benchmarks, followed by an analysis of benchmark results to provide insights into the impact of different design choices on model performance. Through these contributions, we aim to provide a holistic perspective on time-series anomaly detection and highlight promising avenues for future investigation.

Dragos-Florian Ristache, Krzysztof Onak

Small-space frequency estimators play a crucial role in a multitude of settings related to both data science and machine learning in the context of big data processing. Many frequency estimators use internal randomness to compress the information about the frequencies of items to a small sketch that can be used to provide estimates. Historically, these types of estimators were designed without considering the danger of invalid estimates caused by their adaptive use. However, this kind of scenario naturally occurs when they are used as a subroutine in a more complicated algorithm or when an adversary maliciously attempts to corrupt estimates. The reason why the classic way of analyzing these types of algorithms does not provide satisfying guarantees is that it often assumes that the queries and updates are independent of the previously generated estimates. In the adaptive setting, each provided estimate has the potential of leaking information about the estimator's internal randomness, which can, in turn, be used to craft queries and updates on which the frequency estimator does not perform well. In this work, we consider one of the simplest such estimators, Count-Min Sketch, and demonstrate both positive and negative results. On the one hand, we show that Count-Min Sketch can be made resistant to adaptive attacks in both the random oracle model, which corresponds to cryptographically hard hash functions, and using universal hash functions if the domain size is in polynomial relationship with with the size of hash tables. On the other hand, we develop techniques for more efficient attacks on Count-Min Sketch than those known previously. First, we introduce value-based attack techniques. They leverage information from queries to Count-Min Sketch beyond whether a collision between different elements occurs. By inserting carefully selected values, we can more efficiently recover the elements that hash to the same bucket as the targeted element, for which we want to manipulate estimates. Second, for a very popular family of universal hash functions, we demonstrate that it is possible to generate simultaneous collisions between two elements in two hash tables at the rate which in general it is possible to generate collisions in one hash table. This allows for efficiently producing more skewed estimates than those proposed by previous works. This is achieved by developing methods that allow for narrowing the range of internal parameters of these hash functions.

Liang Zhang 0042, Tao Long 0002, Yang Liu 0017, Lei Zhang 0066, Laizhong Cui, Qingjiang Shi

Federated learning (FL) enables the training of a global machine learning model among multiple local clients in a collaborative fashion without directly sharing the details of their data. Due to this advantage, it has been utilized in a wide range of applications where privacy is a critical concern and has attracted great attention for graph representation learning (GRL). Despite the offered advances, there still exist two major challenges in the FL for GRL across distributed graph data, including heterogeneity and complementarity. In order to tackle these challenges, a novel personalized federated graph augmentation (PFGA) framework is proposed in this work. Unlike existing techniques, it utilizes generative models as bridges to enable information sharing among clients, thereby facilitating the collaborative training of GRL models. Instead of directly using the generative model trained on each client individually, we aggregate them into the globally generative model to gain a global view of the entire graph, which effectively alleviates the heterogeneity and complementarity issues simultaneously. We formulate the training of the generative and GRL models as a min-max adversarial learning problem and theoretically prove the convergence. Furthermore, the effectiveness of the method is demonstrated using experimental results on six real-world datasets.

Mike Van Ness, Madeleine Udell

Survival analysis is widely used as a technique to model time-to-event data when some data is censored, particularly in healthcare for predicting future patient risk. In such settings, survival models must be both accurate and interpretable so that users (such as doctors) can trust the model and understand model predictions. While most literature focuses on discrimination, interpretability is equally as important. A successful interpretable model should be able to describe how changing each feature impacts the outcome, and should only use a small number of features. In this paper, we present DyS (pronounced "dice''), a new survival analysis model that achieves both strong discrimination and interpretability. DyS is a feature-sparse Generalized Additive Model, combining feature selection and interpretable prediction into one model. While DyS works well for all survival analysis problems, it is particularly useful for large (in n and p) survival datasets such as those commonly found in observational healthcare studies. Empirical studies show that DyS competes with other state-of-the-art machine learning models for survival analysis, while being highly interpretable.

Furkan Cantürk, Taha Varol, Reyhan Aydoğan, Okan Örsan Özener

By examining the patterns of solutions obtained for various instances, one can gain insights into the structure and behavior of combinatorial optimization (CO) problems and develop efficient algorithms for solving them. Machine learning techniques, especially Graph Neural Networks (GNNs), have shown promise in parametrizing and automating this laborious design process. The inductive bias of GNNs allows for learning solutions to mixed-integer programming (MIP) formulations of constrained CO problems with a relational representation of decision variables and constraints. The trained GNNs can be leveraged with primal heuristics to construct high-quality feasible solutions to CO problems quickly. However, current GNN-based end-to-end learning approaches have limitations for scalable training and generalization on larger-scale instances; therefore, they have been mostly evaluated over small-scale instances. Addressing this issue, our study builds on supervised learning of optimal solutions to the downscaled instances of given large-scale CO problems. We introduce several improvements on a recent GNN model for CO to generalize on instances of a larger scale than those used in training. We also propose a two-stage primal heuristic strategy based on uncertainty-quantification to automatically configure how solution search relies on the predicted decision values. Our models can generalize on 16x upscaled instances of commonly benchmarked five CO problems. Unlike the regressive performance of existing GNN-based CO approaches as the scale of problems increases, the CO pipelines using our models offer an incremental performance improvement relative to CPLEX. The proposed uncertainty-based primal heuristics provide 6-75% better optimality gap values and 45-99% better primal gap values for the 16x upscaled instances and brings immense speedup to obtain high-quality solutions. All these gains are achieved through a computationally efficient modeling approach without sacrificing solution quality.

Da-Wei Zhou, Han-Jia Ye

Machine learning has rapidly progressed, resulting in a vast repository of both general and specialized models that address diverse practical needs. Reusing pre-trained models (PTMs) from public model zoos has emerged as an effective strategy, leveraging rich model resources and reshaping traditional machine learning workflows. These PTMs encapsulate valuable inductive biases beneficial for downstream tasks. Well-designed reuse strategies enable models to be adapted beyond their original scope, enhancing both performance and efficiency in target machine learning systems. This survey offers a unifying perspective on model reuse, establishing connections across various domains and presenting a novel taxonomy that encompasses the full lifecycle of PTM utilization---including selection from model zoos, adaptation techniques, and related areas such as model representation learning. We delve into the similarities and distinctions between reusing specialized and general PTMs, providing insights into their respective advantages and limitations. Furthermore, we discuss key challenges, emerging trends, and future directions in model reuse, aiming to guide research and practice in the era of large-scale pre-trained models. A comprehensive list of papers about model reuse is available at https://github.com/LAMDA-Model-Reuse/Awesome-Model-Reuse.

Xiao-Wen Yang, Jie-Jing Shao, Lan-Zhe Guo, Bo-Wen Zhang, Zhi Zhou, Lin-Han Jia, Wang-Zhou Dai, Yu-Feng Li

Large Language Models (LLMs) have shown promising results across various tasks, yet their reasoning capabilities remain a fundamental challenge. Developing AI systems with strong reasoning capabilities is regarded as a crucial milestone in the pursuit of Artificial General Intelligence (AGI) and has garnered considerable attention from both academia and industry. Various techniques have been explored to enhance the reasoning capabilities of LLMs, with neuro-symbolic approaches being a particularly promising way. This paper comprehensively reviews recent developments in neuro-symbolic approaches for enhancing LLM reasoning. We first present a formalization of reasoning tasks and give a brief introduction to the neuro-symbolic learning paradigm. Then, we discuss neuro-symbolic methods for improving the reasoning capabilities of LLMs from three perspectives: Symbolic->LLM, LLM->Symbolic, and LLM+Symbolic. Finally, we discuss several key challenges and promising future directions. We have also released a GitHub repository including papers and resources related to this survey: https://github.com/LAMDASZ-ML/Awesome-LLM-Reasoning-with-NeSy.